Relationship Scoring Model
Massachusetts Analytics developed a client relationship scoring model used by financial service providers. The system quantifies coverage quality and pinpoints areas of potential relationship improvement, helping users identify sales opportunities and resolve coverage deficiencies.
A global consulting firm wanted to create a data-driven relationship scoring model using its large store of proprietary relationship performance data (quantitative and qualitative). The system needed to do more than simply repackage and display the collected performance data. It had to provide actionable feedback, allowing the financial service providers to identify opportunities and indicate specific ways to improve on weak relationships.
Modeling challenges included:
- Inconsistent subjective scoring in the data
- Dataset consisting of both categorical and continuous variables
- Tight time constraints
Massachusetts Analytics developed a relationship scoring algorithm that accommodates subjectivity in the data and is not adversely influenced by missing data. The system quantifies coverage quality separately from simple metrics such as share of wallet or business volume. The model not only conveys a snapshot of the current relationship, but accounts for trends in relationship quality. The algorithm was designed to be transparent so that it can incorporate expert judgement, and clearly indicates actions that can be taken to improve the user’s coverage model.